OHDSI Home | Forums | Wiki | Github

How to store discription of CT scan result in CDM

cdm

(caihaoyan) #1

for example, the discription of CT scan is " Low-density shadow in frontal lobe. the fourth ventricle enlarged", how to store the information in measurement table? thanks.


(Seng Chan You) #2

@caihaoyan I think you mean ‘report’ of the CT.
I recommend you to use NOTE table for the report linked to the record for CT (measurement).
You can link Measurement and NOTE table by using ‘note_event_id (put ‘Measurement’)’, and note_event_field_concept_id (put ‘measuremernt_id’ here)

Please see this link


(caihaoyan) #3

thanks for the reply. very helpful!
and we have done NLP for the CT report, using entities “location (such as frontal lobe or fourth ventricle)” & “findings (such as low density & enlarged”. could you please further suggest us how to store these informaiton in CDM? thanks again.


(Qi Yang) #4

@caihaoyan

I am not able to find the relevant concepts on “Low-density shadow in frontal lobe” or “the fourth ventricle enlarged”. I think our vocabulary is not detailed to that level of granularity when it comes to radiology reports. In these cases, I would use some higher level concepts such as these:

4236310 Computed tomography of brain abnormal
or
4027563 Decreased radiologic density (for “Low-density shadow in frontal lobe”)

These concepts belong to condition domain and should be loaded in condition table with condition_type_concept_id as 32424 (NLP derived). Then you can use fact_relationship table to link this record with the record in Note (or Note_NLP) table.

Alternatively, you can use measurement table. For example, for measurement_concept_id, use 3004287 (CT Head), or 3002086 (CT Head W contrast IV), for value_as_concept_id, use 45878745 (Abnormal), and for measurement_type_concept_id, use 32423 (NLP derived)

I also talked to severeal veterans in EHR working group on the subject. They recommend using Note_NLP table as it will display your result in a more structured way. Since I am not familiar with that table, I won’t be able to advise you here. I suggest you ask someone in NLP work group.

Hope this helps.


t